Thermal Hall conductivity from semiclassical spin dynamics simulations: implementation and applications to chiral ferromagnets and Kitaev magnets

This paper presents a semiclassical spin dynamics framework for computing thermal Hall conductivity via linear response theory, demonstrating its effectiveness in capturing non-linear magnon interaction effects in chiral ferromagnets and Kitaev magnets beyond simple non-interacting approximations.

The Gist

Imagine you have a crowded dance floor filled with dancers (the spins in a magnet). Usually, we think of heat moving through a material like a crowd of people shuffling from a hot room to a cold one. But in certain special magnetic materials, something weird happens: if you heat one side, the "heat dance" doesn't just move straight across; it starts to swirl sideways, creating a Thermal Hall Effect. It's like if you pushed a crowd of people from the left, and instead of moving right, they started spinning in a circle to the right.

This paper is a guidebook and a report card for a new way of simulating this swirling heat dance using computers.

Here is the breakdown of what the authors did, using simple analogies:

1. The Problem: The "Perfect World" vs. Reality

For a long time, scientists tried to predict this swirling heat using a "Perfect World" theory (called Linear Spin-Wave Theory).

• The Analogy: Imagine trying to predict traffic flow by assuming every car is a ghost that never crashes, never slows down, and never talks to other cars. You draw a perfect map, and the math says, "Okay, traffic will flow this way."
• The Reality: In the real world, cars crash, drivers get distracted, and traffic jams happen. In magnets, the "cars" are heat-carrying particles (magnons). They bump into each other, interact, and get messy, especially when the material is hot. The "Perfect World" math often fails to match what we see in real experiments because it ignores these messy interactions.

2. The Solution: The "Semiclassical Simulation"

The authors built a new computer program to simulate these magnets. Instead of assuming the particles are perfect ghosts, they let them be real, messy dancers.

• The Analogy: Instead of a perfect map, they built a massive virtual dance floor. They put thousands of dancers on it and let them move according to the laws of physics. They didn't tell the dancers how to interact; they just set the rules (the magnetic forces) and let the chaos happen naturally.
• The "Semiclassical" part: This means they treated the dancers as if they were classical objects (like billiard balls) but with some quantum rules attached. It's a middle ground that is fast enough to run on a computer but detailed enough to catch the "messy" interactions that the perfect theory misses.

3. The Two Parts of the Swirl

The paper explains that the total "swirl" (Thermal Hall Conductivity) is actually made of two different things happening at the same time. The authors had to figure out how to measure both separately and then add them up.

• Part A: The "Kubo" Term (The Active Dance):

  • Analogy: This is the energy of the dancers actually moving across the floor because of a temperature difference. It's the active traffic flow.
  • How they measured it: They had to watch the dancers over time, recording how their movements correlated. It's like filming the dance floor for hours and analyzing the video frame-by-frame to see how the crowd moves. This is very hard to calculate because you need a huge amount of data to get a clear picture.

• Part B: The "Energy Magnetization" Term (The Static Spin):

  • Analogy: Even if the dancers aren't moving across the floor, they might be spinning in place in a specific direction. If you push the whole floor, these stationary spins create a tiny, hidden current. It's like a stationary fan blowing air; the fan isn't moving, but it's still pushing air.
  • The Catch: The authors found that if you only measure the "Active Dance" (Part A), you get a wrong answer. You must add the "Static Spin" (Part B) to get the true total. In fact, these two numbers are often huge but have opposite signs, so they almost cancel each other out. Getting the final answer right requires extreme precision, like balancing two giant weights on a scale.

4. What They Tested

They tested their method on two types of "dance floors":

1. A Square Dance (Chiral Ferromagnet): They checked if their computer code worked by comparing it to a known dance routine. It matched perfectly (mostly), proving their method is reliable.
2. The Kitaev Honeycomb (The Tricky One): This is a famous, complex magnetic model that scientists think might host "fractionalized" particles (particles that split in half). They wanted to see if their messy simulation could predict the heat swirl in this complex system.

5. The Big Discovery

When they ran the simulation on the tricky Kitaev model, they found something surprising:

• The "Perfect World" theory failed. It predicted that the heat swirl would get stronger as the material got hotter.
• The Simulation showed the opposite. The swirl got weaker as it got hotter.
• Why? Because the "messy interactions" (the dancers bumping into each other) became too strong. The heat carriers lost their identity and stopped being efficient at carrying heat sideways.

6. Why This Matters

This paper is like a new toolkit for physicists.

• Before: Scientists had to guess how heat moves in complex magnets, often relying on simplified theories that ignored the "noise" of the real world.
• Now: They have a robust, unbiased computer method that can handle the noise, the chaos, and the interactions. It acts as a "benchmark" (a gold standard) to compare against real-world experiments.

In a nutshell:
The authors built a super-accurate computer simulator that treats magnetic heat like a chaotic dance party rather than a perfect parade. They proved that to understand how heat swirls in magnets, you can't just look at the perfect rules; you have to watch the messy, real-time interactions. Their method successfully explains why heat behaves differently in complex magnets than simple theories predict, helping us understand exotic materials that could one day power new technologies.

Technical Summary

Here is a detailed technical summary of the paper "Thermal Hall conductivity from semiclassical spin dynamics simulations: implementation and applications to chiral ferromagnets and Kitaev magnets."

1. Problem Statement

The thermal Hall conductivity ($\kappa_{xy}$) is a crucial probe for exotic excitations in quantum materials, such as fractionalized quasiparticles in spin liquids. However, interpreting experimental $\kappa_{xy}$ data is challenging because:

• Complexity of Mechanisms: The signal arises from a combination of intrinsic Berry curvature effects, extrinsic scattering (skew-scattering, side-jump), and phonon contributions.
• Limitations of Analytical Approaches: Standard theoretical descriptions often rely on Linear Spin-Wave Theory (LSWT) assuming non-interacting magnons. This "intrinsic" approximation fails to capture strong thermal fluctuations, magnon-magnon interactions, and non-linear effects that dominate in many regimes (e.g., near phase transitions or in frustrated systems).
• Numerical Challenges: Calculating $\kappa_{xy}$ numerically is difficult because it is a non-equilibrium, finite-temperature dynamical property requiring large system sizes and long time scales, which are often prohibitive for exact quantum methods due to the exponential growth of the Hilbert space.

The paper addresses the need for an unbiased, numerical framework capable of capturing both intrinsic and extrinsic contributions, including non-linear interactions, to benchmark against experiments.

2. Methodology

The authors develop and implement a semiclassical spin dynamics approach based on the Landau-Lifshitz-Gilbert (LLG) equation.

A. Theoretical Framework

• Linear Response Theory: The thermal Hall conductivity is decomposed into two distinct terms based on the Kubo formula and energy magnetization:
  $$\kappa_{xy} = \kappa_{xy}^{\text{Kubo}} + \kappa_{xy}^{\text{EM}}$$
  • $\kappa_{xy}^{\text{Kubo}}$: Arises from the out-of-equilibrium response to a thermal gradient, calculated via time-correlation functions of energy currents ($\langle I_x(t)I_y(0) \rangle$).
  • $\kappa_{xy}^{\text{EM}}$: Arises from equilibrium energy magnetization currents (circulating currents present even without a gradient).
• Local Energy Currents: The authors derive explicit expressions for local bond energy currents ($j_{j \to i}$) for general spin models using Poisson brackets of local energy densities. This allows the calculation of currents directly from spin configurations without assuming specific quasiparticle forms.
• Classical Limit: In the classical limit ($\hbar/T \to 0$), quantum Kubo correlations coincide with classical correlations, justifying the use of classical spin dynamics.

B. Numerical Implementation

• Dynamics: The system evolves via the stochastic LLG equation.
  • Equilibration Phase: Uses Langevin dynamics (with damping $\alpha_G$ and thermal noise) to reach canonical equilibrium at temperature $T$.
  • Evolution Phase: Uses Hamiltonian dynamics (damping $\alpha_G = 0$) to compute time-correlation functions for the Kubo term, ensuring energy conservation.
• Data Extraction:
  • Kubo Term: Computed by integrating the transverse current-current correlation function $C_{xy}(t)$. The raw data is fitted to damped sine functions to extract the integral and reduce statistical noise.
  • Magnetization Term: Computed via static averages of local currents (real-space method) or via momentum-space formulations.
• Validation: The method is benchmarked against a chiral ferromagnet model on a square lattice (comparing with Ref. [19]) before being applied to the more complex Kitaev model.

3. Key Contributions

1. Robust Numerical Protocol: The paper provides a detailed, pedagogical guide on how to extract $\kappa_{xy}$ from classical spin dynamics, explicitly handling the separation of Kubo and magnetization terms and the technical challenges of time-correlation fitting.
2. Unbiased Approach: Unlike LSWT, this method does not assume well-defined quasiparticles. It naturally incorporates many-body effects, thermal fluctuations, and non-linear interactions on an equal footing.
3. Resolution of Divergences: The authors demonstrate that while $\kappa_{xy}^{\text{Kubo}}$ and $\kappa_{xy}^{\text{EM}}$ individually diverge as $1/T$ at low temperatures (a known artifact of classical statistics for gapless bosons), their sum remains finite and physically meaningful. They establish a systematic procedure to ensure this cancellation.
4. Critique of Equilibrium-Only Methods: The paper refutes the validity of extracting $\kappa_{xy}$ solely from equilibrium edge currents (a method proposed in recent literature for fermionic systems), showing that for bosonic/magnonic systems, neglecting the Kubo term leads to unphysical divergences.

4. Results

A. Benchmark: Chiral Ferromagnet (Square Lattice)

• The method successfully reproduces the behavior of a chiral ferromagnet with Dzyaloshinskii-Moriya interaction (DMI).
• Findings: The total $\kappa_{xy}$ shows a peak near the skyrmion phase transition. The Kubo term exhibits a sharp sign reversal at the transition, while the magnetization term remains smooth. This highlights that dynamical correlations (Kubo) are more sensitive to changes in spin chirality than static averages.

B. Application: Antiferromagnetic Kitaev Model (Honeycomb Lattice)

• System: Investigated in a magnetic field ($B/J = 2.1$) just above the saturation field, a regime relevant to materials like $\alpha$-RuCl$_3$.
• Temperature Dependence:
  • Low $T$: $\kappa_{xy}$ shows a peak. The authors attribute the finite value at very low $T$ (where LSWT predicts activation) to the classical statistics of the simulation (Rayleigh-Jeans distribution), which overpopulates low-energy modes compared to Bose-Einstein statistics.
  • High $T$: $\kappa_{xy}$ decreases roughly as $1/T$.
• Comparison with LSWT:
  • The intrinsic LSWT prediction (non-interacting magnons) predicts a monotonic increase with temperature.
  • The semiclassical simulation shows a non-monotonic, decreasing trend at higher temperatures.
  • Conclusion: The deviation is not just due to statistics but due to strong magnon-magnon interactions. As temperature rises, magnons lose coherence (broadening of spectral peaks), and the quasiparticle picture breaks down. The semiclassical method captures this suppression of transport efficiency, which LSWT misses.
• Longitudinal Conductivity: The longitudinal thermal conductivity $\kappa_{xx}$ follows a power law $\kappa_{xx} \sim T^{-0.68}$, suggesting Umklapp (magnon-magnon) scattering is the dominant relaxation mechanism even at low temperatures.

5. Significance

• Beyond Harmonic Approximation: The work proves that semiclassical spin dynamics is a powerful tool for studying thermal transport in regimes where the harmonic approximation (LSWT) fails, specifically capturing the effects of strong thermal fluctuations and non-linear interactions.
• Experimental Relevance: The results provide a realistic benchmark for interpreting thermal Hall experiments in quantum magnets (like Kitaev candidates), suggesting that observed deviations from intrinsic predictions are likely due to many-body effects rather than just fractionalization or phonon contamination.
• Methodological Standard: By establishing a rigorous protocol for separating and calculating the Kubo and magnetization terms, the paper sets a standard for future numerical studies of thermal transport in magnetic systems, applicable to both ordered and spin-liquid phases.
• Future Directions: The authors suggest extending this framework to include colored noise (to better mimic Bose statistics at low $T$) and coupled spin-phonon systems, further bridging the gap between theory and experiment.